Ben Fritz
👤 SpeakerAppearances Over Time
Podcast Appearances
Und sie haben bemerkt, dass es besser ist, auf der Seite davon zu gehen, als auf der Seite davon zu sagen, ich weiß es nicht. Es gibt einige Versuche, das zu tun. Ich dachte, das Interessanteste war, dass AIs auf der Grundlage der Trainingsdaten gebaut sind und dass sie mehr Informationen einsetzen müssen, training data that includes people not knowing things.
Usually they're learning from data, from facts, things people know, people talking about stuff they know, writing about stuff they know. So can you find a way in the earliest stages to introduce this concept to a new AI? Of course, it's complicated.
Usually they're learning from data, from facts, things people know, people talking about stuff they know, writing about stuff they know. So can you find a way in the earliest stages to introduce this concept to a new AI? Of course, it's complicated.
If you teach it, I don't know too much, then it's going to start doubting itself and it's going to not come up with answers that it actually could come up with. But clearly right now, a growing number of people think it's going too far in the other direction.
If you teach it, I don't know too much, then it's going to start doubting itself and it's going to not come up with answers that it actually could come up with. But clearly right now, a growing number of people think it's going too far in the other direction.